Identification of Nutritional Patterns in Maize By the Images Analysis.

Image analysis is a technique capable of providing information extracted from the leaves, which may contribute to the early identification of nutrient deficiency. The objective was to evaluate nutritional patterns of nitrogen (N) and potassium (K) in maize plants subjected to combined levels of the both fertilizers, and their effects on the yield, by the image analysis. The study was conducted in no-tillage system area. The experimental design was a randomized block in a factorial design with 6 N doses (0, 60, 120, 180, 240 and 300 kg.ha^-1) combined with 4 doses of K (0, 30, 60 and 90 kg. ha^-1) with 3 replicates, which were established after the V4 stage (topdressing of corn). Samples of maize leaves were taken (6 of each plot) in V6 and R1 stage for N and K analysis. At the end of the cycle the yield was assessed. The leaves in V6 stage were digitalized at 1200 dpi and data was analyzed the Matlab software. In the images were extracted blocks of 20*20 pixel and of them extracted characteristics of 4 spectral indices. After statistical classification were obtained the global percentage of right (GPR) and the Kappa index. The results for the nutrient content in leaves and yield were statistically analyzed through regression study. The Level of 235 kg.ha^-1 provided the highest yield (8.8 t.ha^-1), and 185 kg.ha^-1 the maximum N content in the leaves (35.75 g.kg^-1) at V6; at R1 was observed linear effect for N content. The image analysis was able to identify to levels N*K 0/0, 120/0, 130/30, 180/30, 240/30, 0/60, 240/60, 300/60, 0/90 and 240/90 with GPR of 85% and excellent Kappa index (0.8) when combined with three or more spectral indices. Spectral indices enabled the identification of nutritional patterns of N and K at V6 stage of maize.